Hamel Husain uses Claude as his main assistant, Codex as a code reviewer that feeds back into Claude, and Gemini for video, PDFs, and audio. Custom plugins let Claude call Gemini automatically and send code to Codex for Google-style review, keeping everything in one workflow.
Which AI should you use? Claude, OpenAI, or Gemini? They each have their own strengths, which means you should use all of them. Here's what I do. Claude is my daily driver for coding and knowledge work. It writes code like a startup engineer, and it's super pragmatic and just ships. And it has the best first-party plugins. I use a Chrome extension that controls my browser, which is useful for sites that require authentication. I use OpenAI's Codex as my code reviewer. It develops more like a Google engineer. Thorough and rigorous and sometimes loves to over-engineer. I have it review everything Claude writes, and then Claude reads the feedback and pushes back where it disagrees. I built a plugin that runs this back and forth for you. For anything that isn't text, like videos, PDFs, and audio, I use Gemini. Gemini is simply the best for these other formats. I have another plugin that lets Claude call Gemini whenever it needs to see this kind of format. No copying and pasting between tools. By linking everything together, together with a plugin, you get the best of each while staying in Claude the whole time. Links to all the plugins are in the comments so you can try this yourself.
